Aug 7, 2026 · 8 min · 8 segments
In June 2026, the last of the top-eight Source-to-Pay suites on Gartner's Magic Quadrant quietly shipped its "AI Agent Studio". Now, every serious procurement platform has one. But before you take…
Joël Collin-DemersHostOkay, so let's define this thing because the term is slippery and honestly, that's half the challenge.
An AI agent studio is a no-code or low-code environment built right into your enterprise platform where your team designs and wires up LLM-supported business processes.
So instead of hard coding one fixed word flow the way we've done it for years, you drop LLM-powered steps into a process and you chain them together towards an outcome that you're looking for.
It's a place where you can build these processes and where governance controls decides what the model gets to act on, where a plain boring deterministic algorithm does the job instead, and where a human has to stay in the loop.
So the skill worth building is recognizing the thing no matter what the vendor stamps on the box.
There's a real difference between a studio and embedded AI, and most vendors are going to blur that line in the demo.
Nearly every platform now ships with LLM features baked straight into their workflows.
Those features are useful, but they're the vendor's processes running the vendor's logic, not yours.
That shift from renting a fixed LLM supported process to building your own is the real story underneath all the naming noise.
Okay, so let's define this thing because the term is slippery and honestly, that's half the challenge.
An AI agent studio is a no-code or low-code environment built right into your enterprise platform where your team designs and wires up LLM-supported business processes.
So instead of hard coding one fixed word flow the way we've done it for years, you drop LLM-powered steps into a process and you chain them together towards an outcome that you're looking for.
It's a place where you can build these processes and where governance controls decides what the model gets to act on, where a plain boring deterministic algorithm does the job instead, and where a human has to stay in the loop.
So the skill worth building is recognizing the thing no matter what the vendor stamps on the box.
There's a real difference between a studio and embedded AI, and most vendors are going to blur that line in the demo.
Nearly every platform now ships with LLM features baked straight into their workflows.
Those features are useful, but they're the vendor's processes running the vendor's logic, not yours.
That shift from renting a fixed LLM supported process to building your own is the real story underneath all the naming noise.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
